Extended Emission from Short Gamma-Ray Bursts Detected with SPI-ACS/INTEGRAL

The short duration (T90 < 2 s) gamma-ray bursts (GRBs) detected in the SPI-ACS experiment onboard the INTEGRAL observatory are investigated. Averaged light curves have been constructed for various groups of events, including short GRBs and unidentified short events. Extended emission has been found in the averaged light curves of both short GRBs and unidentified short events. It is shown that the fraction of the short GRBs in the total number of SPI-ACS GRBs can range from 30 to 45%, which is considerably larger than has been thought previously.

💡 Research Summary

This paper investigates the presence of extended emission (EE) in short gamma‑ray bursts (GRBs) using data from the SPI‑ACS anticoincidence shield aboard the INTEGRAL observatory. The authors compiled a catalog of 83 confirmed short GRBs (T90 < 2 s) detected between 2002 and 2007, extending the earlier Rau et al. (2005) list. From these, 53 light curves were available for detailed analysis. In addition, they examined 105 short‑duration events (T90 < 2 s) that were not confirmed by other missions, of which 43 had extremely brief durations (T90 ≤ 0.05 s) and are suspected to be particle‑induced triggers. A fourth test group of 33 events known to be particle triggers was also studied for comparison.

SPI‑ACS records the total count rate of its BGO scintillators with a 50 ms time resolution and a lower energy threshold of ≈ 80 keV, providing an almost all‑sky view but no spectral information. To isolate genuine GRB signals, the authors applied a multi‑step selection: they excluded events coincident with the INTEGRAL Radiation Monitor (IREM) or GOES particle alerts, removed soft‑gamma repeater (SGR) bursts, and required a significance > 12 σ in the native 50 ms bins.

For each event the light curve was extracted from –150 s to +200 s relative to the trigger peak. Background was modeled linearly using the intervals –150 → –50 s and +100 → +200 s, which were shown to be stable to within 0.3 % over the 350 s timescale. The authors noted that the statistical variance of the count rate exceeds pure Poisson expectations, following a 1.57 × √B law, and incorporated this into their σ calculations. After background subtraction, the data were rebinned progressively from 0.05 s up to 5 s, and all light curves within a given group were aligned on the main peak and averaged point‑by‑point.

The averaged light curves reveal a statistically significant EE component lasting ≈ 25 s after the initial short spike for both the confirmed short GRBs (first group) and the unconfirmed short events (second group). The EE in the confirmed sample has an average count‑rate of 46 ± 15 counts s⁻¹, while the EE amplitude in the unconfirmed sample suggests that roughly 84 ± 35 % of those events are genuine GRBs rather than particle triggers. In contrast, the test groups (the very short T90 ≤ 0.05 s events and the known particle‑trigger sample) show no EE, reinforcing the interpretation that EE is an intrinsic astrophysical feature.

Comparing with other missions, the authors note that similar EE has been reported in BATSE, Konus‑Wind, and BeppoSAX data, indicating that EE is a common property of short GRBs across a wide energy range. Because SPI‑ACS has a higher low‑energy threshold (≈ 80 keV) than BATSE (≈ 50 keV) and a comparable trigger window (50 ms vs. 64 ms), it preferentially detects the harder subset of short GRBs. Consequently, the fraction of short GRBs among all SPI‑ACS detections is estimated to lie between 30 % and 45 %, substantially higher than the ≈ 25 % fraction reported for BATSE. This suggests that a sizable population of hard short GRBs has been under‑represented in earlier catalogs.

The paper discusses the physical origin of EE, debating whether it represents prolonged central‑engine activity or the onset of an X‑ray afterglow. The lack of spectral lag in most EE‑bearing bursts (e.g., GRB 060614) argues against a simple long‑GRB classification, while the softer spectrum of EE relative to the initial spike aligns with afterglow expectations. The authors advocate for multi‑wavelength follow‑up, spectral lag measurements, and placement on the Amati Eₚ–E_iso diagram to disentangle these possibilities.

In conclusion, the study provides robust statistical evidence for extended emission in the averaged light curves of short GRBs detected by SPI‑ACS, demonstrates that a large fraction of unconfirmed short events are likely real GRBs, and revises upward the estimated proportion of short GRBs in the INTEGRAL/ACS sample. These findings have implications for GRB classification schemes, the demographics of compact‑object mergers, and the design of future high‑energy transient monitors.